Management of medication using a mobile robot and artificial intelligence

摘要

It is important to take a pill in time to maximize effect of the treatment. However, people often forget to take a pill or they are confused whether they took it or not. In this paper, we propose a method to manage medication of elderly people by using a mobile robot and artificial intelligence. When it is time for elderly person to take a pill, a mobile robot comes to him by employing ultrawideband (UWB)-based robot navigation and suggests to take a pill. Then, the robot starts to collect 3D skeleton data by using a depth sensor installed in the robot. At last, it recognizes an action of taking a pill by using the collected 3D skeleton data and gated recurrent unit (GRU)-based classifier. We used the distances between upper joints instead of the raw skeleton data to obtain robustness for location and standing direction of the human. The distances were fed to the GRU-based binary classifier for the decision. As a result, we achieved the recognition accuracy of 95.33%.